Abstract In this work, we present a novel methodology for decision trees that use oversampling, not before tree construction (in the entire dataset), but inside each internal node (and corresponding input space region) of the tree. This strategy proves to be successful in fighting the greedy nature of decision trees. We take also into consideration the nature of the input variables, not just quantitative or binary, and also introduce the use of novel distances between instances that can also be used in other contexts. The application of our methodology to a significant number of datasets, thirteen, both balanced and imbalanced problems, shows the relevance of our approach when compared to CART and C5.0. Although our experiments were conducted on a standard computing platform, the proposed approach is well suited for high-performance computing environments, since node-level oversampling and distance computations can be efficiently parallelized, enabling the method to scale to large and high-dimensional datasets.

Convolutional neural networks are shown to be vulnerable to changes in the background. The proposed method is an end-to-end method that augments the training set by introducing new backgrounds during the training process. These backgrounds are created by a generative network that is trained as an adversary to the model. A case study is explored based on overhead power line insulators detection using a drone - a training set is prepared from photographs taken inside a laboratory and then evaluated using photographs that are harder to collect from outside the laboratory. The proposed method improves performance by over 20% for this case study.

In classification, when there is a disproportion in the number of observations in each class, the data is said to be class imbalance. Class imbalance is pervasive in real world applications of data classification and has been the focus of much research. The minority class contributes too little to the decision boundary because the learning process learns from each observation in isolation. In this paper, we discuss the application of learning pairwise rankers as a solution to class imbalance. We compare ranking models to alternatives from the literature.

In this work, we introduce in the first part new developments in Principal Component Analysis (PCA) and in the second part a new method to select variables (genes in our application). Our focus is on problems where the values taken by each variable do not all have the same importance and where the data may be contaminated with noise and contain outliers, as is the case with microarray data. The usual PCA is not appropriate to deal with this kind of problems. In this context, we propose the use of a new correlation coefficient as an alternative to Pearson's. This leads to a so-called weighted PCA (WPCA). In order to illustrate the features of our WPCA and compare it with the usual PCA, we consider the problem of analyzing gene expression data sets. In the second part of this work, we propose a new PCA-based algorithm to iteratively select the most important genes in a microarray data set. We show that this algorithm produces better results when our WPCA is used instead of the usual PCA. Furthermore, by using Support Vector Machines, we show that it can compete with the Significance Analysis of Microarrays algorithm.

Support vector machines (SVMs) were initially proposed to solve problems with two classes. Despite the myriad of schemes for multiclassification with SVMs proposed since then, little work has been done for the case where the classes are ordered. Usually one constructs a nominal classifier and a posteriori defines the order. The definition of an ordinal classifier leads to a better generalisation. Moreover, most of the techniques presented so far in the literature can generate ambiguous regions. All-at-Once methods have been proposed to solve this issue. In this work we devise a new SVM methodology based on the unimodal paradigm with the All-at-Once scheme for the ordinal classification.

Breast cancer conservative treatment (BCCT) is considered the gold standard of breast cancer treatment. However, aesthetic results are heterogeneous and difficult to evaluate in a standardised way. The limited reproducibility of subjective aesthetic evaluation in BCCT forced the research on objective methods. A recent computer system was developed to objectively and automatically evaluate the aesthetic result of BCCT. In this system, the detection of the breast contour on the digital photograph of the patient is necessary to extract the features subsequently used in the evaluation process. In this paper we extend an algorithm based on the shortest path on a graph to detect automatically the breast contour. The advantage of graph algorithms is that they are guaranteed to find the global optimum of the problem; the difficulty is that they make it hard to enforce shape constraints. We define and compare different techniques to introduce the a priory knowledge of the mammary contour. Experimental results show that the proposed techniques consistently outperform the base method.

Cosmetic assessment of conservative breast cancer treatment plays a major role in the study of breast cancer techniques. Objective assessment methods are being preferred to overcome the drawbacks of subjective evaluation. In this paper a methodology for the objective assessment of conservative breast cancer treatment is proposed. The quantitative measures used in this research provide an objective way to calculate the overall cosmetic result. We report experiments using support vector machines to derive an optimal assessment rule. The results seem to indicate that it is possible to construct an algorithm for a complete objective classification of the aesthetic result of breast conservative treatment.